Mining Product Recommendation from Query Reformulations

1. A method comprising: receiving a query, from a device of a user, that corresponds to a particular product; accessing, from data storage, transition data for the particular product, the transition data including values indicating a number of search transitions based on a change in queries between the particular product and other products in previous session sets of queries, the transition data determined based on analysis of the previous session sets of queries; determining, using a hardware processor, one or more recommendations based on the transition data, the determining including removing a particular recommendation based on a threshold number of non-selections of the particular recommendation by the user; and providing the one or more recommendations to the device of the user.

2. The method of claim 1 , wherein the determining comprises applying a weighting function to the transition data based on a recommendation depth, the recommendation depth being a level corresponding to a number of serial transitions between the particular product and each product of the one or more recommendations.

3. The method of claim 1 , wherein the determining comprises applying a weighting function for transition data at a recommendation depth level greater than one.

4. The method of claim 1 , further comprising generating the transition data by incrementing a transition count corresponding to the particular product and each product of the one or more recommendations each time a search query changes from the particular product to each product of the one or more recommendations in a same search session of the previous session sets of queries.

5. The method of claim 1 , wherein the determining comprises using one-way transition data between the particular product and each product of the one or more recommendations.

6. The method of claim 1 , wherein the determining comprises using a weighted function of one-way transition data between the particular product and each product of the one or more recommendations and one-way transition data between each product of the one or more recommendations and the particular product.

7. The method of claim 1 , wherein the one or more recommendations comprise a product in a complementary category of a category of the particular product.

8. The method of claim 1 , wherein the particular product is published information.

9. The method of claim 1 , further comprising, based on the user selecting a recommendation, increasing a transition count of the transition data for a link between the particular product and a product of the selected recommendation by a weighted incremental value.

10. The method of claim 1 , further comprising, based on the user not selecting a recommendation, decrementing a transition count of the transition data for a link between the particular product and a product of the non-selected recommendation.

11. A system comprising: one or more hardware processors configured to: receive a query, from a device of a user, that corresponds to a particular product; access, from data storage, transition data for the particular product, the transition data including values indicating a number of search transitions based on a change in queries between the particular product and other products in previous session sets of queries, the transition data determined based on analysis of the previous session sets of queries; determine one or more recommendations based on the transition data, the one or more recommendations determined, in part, by removing a particular recommendation based on a threshold number of non-selections of the particular recommendation by the user; and provide the one or more recommendations to the device of the user.

12. The system of claim 11 , wherein the one or more hardware processors are further to generate the transition data by incrementing a transition count corresponding to the particular product and each product of the one or more recommendations each time a search query changes from the particular product to each product of the one or more recommendations in a same search session of the previous session sets of queries.

13. The system of claim 11 , wherein the one or more hardware processors are further to, based on the user selecting a recommendation, increase a transition count of the transition data for a link between the particular product and a product of the selected recommendation by a weighted incremental value.

14. The system of claim 11 , wherein the one or more hardware processors are further to, based on the user not selecting a recommendation, decrement a transition count of the transition data for a link between the particular product and a product of the non-selected recommendation.

15. A machine-readable medium having no transitory signals and storing instructions which, when executed by the at least one processor of a machine, cause the machine to perform operations comprising: receiving a query, from a device of a user, that corresponds to a particular product; accessing, from data storage, transition data for the particular product, the transition data including values indicating a number of search transitions based on a change in queries between the particular product and other products in previous session sets of queries, the transition data determined based on analysis of the previous session sets of queries; determining one or more recommendations based on the transition data, the determining including removing a particular recommendation based on a threshold number of non-selections of the particular recommendation by the user; and providing the one or more recommendations to the device of the user.

16. The machine-readable medium of claim 15 , wherein the determining comprises applying a weighting function to the transition data based on a recommendation depth, the recommendation depth being a level corresponding to a number of serial transitions between the particular product and each product of the one or more recommendations.

17. The machine-readable medium of claim 15 , wherein the determining comprises using one-way transition data between the particular product and each product of the one or more recommendations.

18. The machine-readable medium of claim 15 , wherein the determining comprises using a weighted function of one-way transition data between the particular product and each product of the one or more recommendations and one-way transition data between each product of the one or more recommendations and the particular product.

19. The machine-readable medium of claim 15 , wherein the operations further comprise, based on the user selecting a recommendation, increasing a transition count of the transition data for a link between the particular product and a product of the selected recommendation by a weighted incremental value.

20. The machine-readable medium of claim 15 , wherein the operations further comprise, based on the user not selecting a recommendation, decrementing a transition count of the transition data for a link between the particular product and a product of the non-selected recommendation.